Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DDIPrompt: Drug-Drug Interaction Event Prediction based on Graph Prompt Learning

Yingying Wang, Yun Xiong, Xixi Wu, Xiangguo Sun, Jiawei Zhang

TL;DR

DDIPrompt features a hierarchical pre-training strategy to foster a generalized and comprehensive understanding of drug properties and implements a prototype-enhanced prompting mechanism during inference that effectively harnesses the rich pre-training knowledge to enhance prediction accuracy, particularly for these rare but crucial interactions.

Abstract

Drug combinations can cause adverse drug-drug interactions(DDIs). Identifying specific effects is crucial for developing safer therapies. Previous works on DDI event prediction have typically been limited to using labels of specific events as supervision, which renders them insufficient to address two significant challenges: (1) the bias caused by \textbf{highly imbalanced event distribution} where certain interaction types are vastly under-represented. (2) the \textbf{scarcity of labeled data for rare events}, a pervasive issue where rare yet potentially critical interactions are often overlooked or under-explored due to limited available data. In response, we offer ``DDIPrompt'', an innovative solution inspired by the recent advancements in graph prompt learning. Our framework aims to address these issues by leveraging the intrinsic knowledge from pre-trained models, which can be efficiently deployed with minimal downstream data. Specifically, to solve the first challenge, DDIPrompt features a hierarchical pre-training strategy to foster a generalized and comprehensive understanding of drug properties. It captures intra-molecular structures through augmented links based on structural proximity between drugs, further learns inter-molecular interactions emphasizing edge connections rather than concrete catagories. For the second challenge, we implement a prototype-enhanced prompting mechanism during inference. This mechanism, refined by few-shot examples from each category, effectively harnesses the rich pre-training knowledge to enhance prediction accuracy, particularly for these rare but crucial interactions. Extensive experiments on two benchmark datasets demonstrate DDIPrompt's SOTA performance, especially for those rare DDI events.

DDIPrompt: Drug-Drug Interaction Event Prediction based on Graph Prompt Learning

TL;DR

DDIPrompt features a hierarchical pre-training strategy to foster a generalized and comprehensive understanding of drug properties and implements a prototype-enhanced prompting mechanism during inference that effectively harnesses the rich pre-training knowledge to enhance prediction accuracy, particularly for these rare but crucial interactions.

Abstract

Drug combinations can cause adverse drug-drug interactions(DDIs). Identifying specific effects is crucial for developing safer therapies. Previous works on DDI event prediction have typically been limited to using labels of specific events as supervision, which renders them insufficient to address two significant challenges: (1) the bias caused by \textbf{highly imbalanced event distribution} where certain interaction types are vastly under-represented. (2) the \textbf{scarcity of labeled data for rare events}, a pervasive issue where rare yet potentially critical interactions are often overlooked or under-explored due to limited available data. In response, we offer ``DDIPrompt'', an innovative solution inspired by the recent advancements in graph prompt learning. Our framework aims to address these issues by leveraging the intrinsic knowledge from pre-trained models, which can be efficiently deployed with minimal downstream data. Specifically, to solve the first challenge, DDIPrompt features a hierarchical pre-training strategy to foster a generalized and comprehensive understanding of drug properties. It captures intra-molecular structures through augmented links based on structural proximity between drugs, further learns inter-molecular interactions emphasizing edge connections rather than concrete catagories. For the second challenge, we implement a prototype-enhanced prompting mechanism during inference. This mechanism, refined by few-shot examples from each category, effectively harnesses the rich pre-training knowledge to enhance prediction accuracy, particularly for these rare but crucial interactions. Extensive experiments on two benchmark datasets demonstrate DDIPrompt's SOTA performance, especially for those rare DDI events.
Paper Structure (28 sections, 6 equations, 5 figures, 4 tables, 3 algorithms)

This paper contains 28 sections, 6 equations, 5 figures, 4 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. GNNs for DDI Event Prediction
  4. Graph Prompt Learning
  5. Methodology
  6. Task Formulation
  7. Hierarchical Pre-training
  8. Pretext 1: Molecule Similarity Score Prediction
  9. Pretext 2: Link Prediction
  10. Prompt-based Learning and Inference
  11. Prompt Initialization
  12. Prompt Tuning
  13. Prompt-assisted Inference
  14. Complexity Analysis
  15. Experiment
  16. ...and 13 more sections

Figures (5)

  • Figure 1: Overview of DDIPrompt Framework. During the pre-training stage, we propose a hierarchical method that first captures structural information of molecules through pairwise similarity prediction and then learns interactive proximity via link prediction. With such a powerful pre-trained model, the prompt learning stage can intelligently finish the downstream prediction task with only a few samples, alleviating previously notorious requirement of labels.
  • Figure 2: Sensitivity Study of Different Methods to Training Sample Size.
  • Figure 3: Efficiency analysis on Ryu's dataset.
  • Figure 4: Hyperparameters study on the number of neighboring drugs in $\text{Molecular-GNN}_{\Theta_1}(\cdot)$. Experimental results on Ryu's dataset are reported.
  • Figure 5: Hyperparameters study on the number of convolutional layers in $\text{DDI-GNN}_{\Theta_2}(\cdot)$. Experimental results on Ryu's dataset are reported.